Candida albicans causes severe infections with high mortality rates especially in immunocompromised patients. The aim of this study was to evaluate the efficiency of randomly amplified polymorphic DNA (RAPD) method and compare the discriminatory powers (DP) of different primers used for genotyping Candida albicans isolates. A total of 109 C. albicans strains recovered from throat, sputum, blood, feces, urine, vagina and wound cultures of 65 hospitalized paediatric patients with haematologic malignancy were evaluated by RAPD method using 10 different primers (OPE-03, OPE-04, OPE-12, OPE-18, OPE-19, OPE-20, OPF-10, OPF-12, P1 and P2) between June 1999-April 2003. Strains were separated into groups by analyzing band patterns derived from each primer and the DP was calculated. Reproducibility of the method was determined by evaluating randomly chosen 20 isolates with the same and different PCR devices under the same PCR conditions. C. albicans isolates generated 1-16 bands and were grouped in 41-80 genotypes depending on the primers used. DP of the RAPD method was calculated as > or = 0.90 for each primer (range between 0.90-0.99), which were accepted as reliable values. However, the strains clustered in the same group when studied with a primer could be dispersed into different groups by another primer. The reproducibility of the method was poor and the comparison of band patterns was difficult especially in isolates which generated many bands. In conclusion, for obtaining reliable results by RAPD method, using more than one primer and comparative analysis of these primers are appropriate. RAPD is an adequate method for studying small outbreaks in which a few number of isolates are evaluated, but it is laborious and unreliable for many number of isolates recovered in a long time period because of its poor reproducibility and difficulties in evaluating the strains generating many bands.